african multidisciplinary journal of research (amjr) vol

African Multidisciplinary Journal of Research (AMJR) Vol. 5 (2), 2020, ISSN

2518-2986 (21- 47)

Impact of Agribusiness Training on Youth Empowerment: A Case Study of Fadama

Graduate Unemployed Youth and Women Support (GUYS) Programme in Nigeria

Adeyanju Dolapo 1

Mburu John 2

International Institute of Tropical Agriculture, Ibadan, Nigeria

University of Nairobi


2518-2986 (21-47)

i

Abstract

Agripreneurship is increasingly being recognized as an important and valuable strategy to reduce

the high dependency of young people on white-collar jobs as well as increase employment

opportunities in the non-formal sector. Thus, it has become one of the key African governments’

investments in the creation of sustainable employment as well as improvement of livelihoods of

young people. Based on this, development stakeholders in many African countries have come

together in recent times to support agripreneurship by organizing agribusiness training

programmes which specifically target this category of the population. Despite the numerous

training interventions, there is a dearth of empirical evidence on what has worked and what has

not. Using the case of Fadama Graduate Unemployed Youth and Women Support (GUYS)

programme, this study used the Propensity Score Matching (PSM) method to investigate the

impact of agribusiness training on youth empowerment in Nigeria. A total of 977 respondents

comprising of 455 participants and 522 non-participants were sampled across three states. PSM

model results showed that after controlling for all confounding factors, participation in training

in the Programme had about 11 percent increase in youth empowerment index score. This

implied a positive change in the economic status and livelihoods of the youths who participated

in the agribusiness training of the Programme. Thus, the study suggests that programmes such as

the Fadama GUYS should be encouraged and out-scaled elsewhere in Africa as they can inspire

youths to engage in agribusiness and thereby contribute to reduction of youth unemployment as

well as enhancement of youth empowerment.

Keywords: Agribusiness, Youth empowerment, Youth unemployment, Agribusiness training


2518-2986 (21-47)

1

Introduction

Nigeria is the most populous country in Africa with a population of about 195,874,740,1 out of

which according to Awogbenle and Iwuamadi (2010), 60% is between 18 and 35 years, an age

category regarded to as youths (Nigeria Youth Policy Document, 2009). Hence, it can be inferred

that Nigeria is largely a youthful country (Emeh & Eke, 2012). The country is enormously

endowed with abundant resources capable of empowering the youths for positive outcomes

which can bring about sustainable social and economic development (Odeh & Okoye, 2014).

Despite many studies proving that youths are valuable assets which is germane to successful and

sustainable nation-building (Mutuku, 2011; Hope, 2012), the general characteristics of Nigeria

youths depicted by a high rate of unemployment and underemployment clearly indicate that this

segment of the population is faced with tough economic challenges. The National Bureau of

Statistics (2018) reported that as at the third-quarter of 2018, 55.4 per cent of young people were

either underemployed or unemployed (doing nothing) compared to 52.6 per cent in the same

period of the previous year (2017). This report shows an increased unemployment rate of about 5

per cent in the space of one year. Over the years, unemployment has eaten really deep into the

fabric of Nigeria thereby pushing many young people to either roam the streets meaninglessly

searching for a living (Ali & Salisu, 2019) or engage in illegal activities and thereby increasing

crime rate in the country. In view of this, stakeholders have come to realize that the most

challenging policy question that has to be addressed sooner than later is the growing rate of

youth unemployment. Hope (2012) affirms that to mitigate the several challenges faced by young

people in Nigeria, government and development partners would have to focus on developing

and implementing relevant and sustainable strategies, policies and programmes which favours

this large segment of the population. According to the author, failure to do this could lead to

enormous political, social, economic, and cultural consequences which is already evident in the

country. It is for these reasons that trainings such as vocational and entrepreneurial training came

into being. This will help to convey relevant skills to young people which will not only reduce

unemployment – a severe national scourge - but will also aid youth empowerment.

Entrepreneurship development has been closely linked to social and economic development.

1 https://data.worldbank.org/indicator/SP.POP.TOTL?locations=NG&view=chart visited on July 25, 2019

https://data.worldbank.org/indicator/SP.POP.TOTL?locations=NG&view=chart


2518-2986 (21-47)

2

It is believed that a lasting solution to youth unemployment is embedded in helping young

people to create their own jobs (become Entrepreneurs). This is well articulated by Maigida, et

al. (2013) who argued that the only sustainable way to empower Nigerian youths is to provide

them with adequate and qualitative training and education which will enable them to be job

creators instead of seekers. At the centre of all these job creation initiatives is the agricultural

sector which has been identified as one of the sectors capable of generating sustainable

employment opportunities for a large number of young people (Yami et al., 2019). Based on this,

agribusiness training has become one of the important Nigerian government initiatives to

empower young people as well as inspire them to become job creators.

In recent times, the government with the support of development partners have shown their

political commitment towards empowering young people through agribusiness training.

According to Yami et al. (2019), there are increasing investment on agricultural programmes

aimed at promoting youth participation in agribusiness to reduce youth unemployment problems

and subsequently empower this category of the population. The country has seen a remarkable

number of interventions by both the private and public sectors in recent times. Examples include;

Npower, Youth Commercial Agriculture Development Programme (YCAD), Youth Employment

in Agriculture Programme (YEAP), Youth Initiatives for Sustainable Agriculture (YISA) and

The Livelihood Improvement Family Enterprise (LIFE). Awogbenle and Iwuamadi (2010)

reported that between 1986 and now, there have been a number of remarkable initiatives by

various administrations to promote youth empowerment through the generation of gainful self-

employment This also corroborates the findings of Yami et al. (2019) that indeed there have

been numerous interventions aimed at achieving youth empowerment through agribusiness in

Nigeria. However, these authors also agreed that despite the increasing number of interventions,

there is a dearth of empirical evidence as to what worked or what did not, making it difficult to

make practical policy recommendations.

It is also worth noting that there is very scanty literature on youth empowerment in Nigeria

(Emeh & Eke, 2012; Okoli & Okoli, 2013; Ibrahim, 2013; Hashim, 2014). To fill this research

gap, this study investigated the impact of agribusiness training on youth empowerment, taking

the case of Fadama Graduate Unemployed Youth and Women Support (GUYS) programme.


2518-2986 (21-47)

3

Thus, the study will serve as a guide to those aiming to research similar subject as well as

provide practical evidence on the impact of youth programmes, thereby, driving evidence-based

policy making which could benefit other African countries.

Background on Fadama GUYS Programme

In an effort to reduce youth unemployment and empower young people through agribusiness, the

Fadama GUYS programme was introduced in 2017 by the Federal government of Nigeria, in

collaboration with the World Bank and state governments. Funding was through a tri-partite

agreement between each state government, the federal government and the World Bank. The

programme covered a total of twenty-three states across Nigeria and targeted young people

between 18 and 35 years of age. The training covered different agribusiness fields which include;

Agricultural production (Crop and livestock), agricultural marketing, crop and livestock

processing and financial management.

Theoretical background

The paper is anchored on the theory of change. According to Rogers (2014), the theory is the

building block for impact evaluation. It is a key which underpins any impact evaluation, given

the cause-and-effect focus of the current study (Gertler et al., 2016). The theory was developed

by Weiss (1995) and it describes how and why an initiative (such as training intervention) works.

In other words, “it explains how the activities undertaken by an intervention” (such as a project,

programme or policy) contributes to the result or the set of results which lead to expected or

observed impacts. According to Gertler et al. (2016), the theory describes a chain of events

which results into outcomes, explore the conditions needed to arrive at the outcome and clearly

shows the causal logic behind the programme.

In the current study, the ultimate goal of the Fadama GUYS programme was to empower young

people through agribusiness. To achieve this, the primary initiative taken was agricultural

training which captures training on animal/crop production, marketing of agricultural products,

processing, and financial management practices. These chains of events are expected to empower

the participants to have better economic outcomes (Gertler et al., 2016). Specifically, participants

are trained in different areas of agribusiness to gain desirable skills and attributes which could


2518-2986 (21-47)

4

contribute to their agripreneurship performance and subsequently empower them in the field of

agribusiness.

Study Methodology

Study Area

The study was conducted in three states across Nigeria between January and March 2019. These

include Abia, Ekiti and Kebbi States representing the South-eastern, South-western and North-

western regions respectively. Abia state occupies a total land area of about 4,900 sq km. The

estimated population as of 2016 was about 3,699,168 people (National Bureau of Statistics,

2011). Ekiti state is mainly an upland zone with a total land area of about 5,435 sq km. As of

2006, the state had a population of 2,398,957 people (National Bureau of Statistics, 2011) of

which more than 75 percent were actively engaged in Agriculture. Kebbi state is located in the

north-western part of Nigeria with a total land area of about 36,985 km sq out of which 12,600

km sq is cultivated for agricultural purposes. According to the National Bureau of Statistics

(2011), the estimated population of Kebbi State as of 2016 was 4,440,000. As indicated by the

United Nations Development Programme (2018), as of 2017, the unemployment rate in these

states were of 39.6 percent, 18.6 percent and 11.6 percent, respectively.

Figure 1: Map of Nigeria showing the study areas


2518-2986 (21-47)

5

Sampling and Data Collection

Primary data was sourced from a total of 977 respondents comprising of 455 Fadama GUYS

programme participants and 522 non-participants in the study areas. The survey included

detailed information on key youth empowerment variables and other relevant socioeconomic

characteristics such as age, gender, education and marital status. The questions were

programmed on Open data kit and data were collected using phones and tablets by trained

enumerators across the three states.

The study adopted a multi-stage sampling technique. In the first stage, three states were

purposively selected. The choice of these states was based on the relatively high number of

participants in the Fadama GUYS programme in 2017, to ensure representation of a state from

each region where the programme was conducted (Northern, Western and Eastern), and

similarity of the states in terms of specific characteristic since the three states ranked high is

agricultural activities (more than 70 per cent of the population in all the states are engaged in

agriculture). The aim of this was to ensure that the respondents are comparable to allow

aggregation and generalization of results. In the second stage, the study population was divided

into two strata: participants and non-participants. In the third stage, respondents were randomly

selected from two sampling frames. The first sampling frame consisting of a complete list of

youths who were trained under the FGP in 2017 was used in gathering the treated group and a

second sampling frame consisting the list of community youths obtained from the local

governments where the training was conducted was used in gathering the control group. The

random selection of both the treatment and control group was done via random numbers

generated using Microsoft Excel. The proposed sample size for participants was 488 and 600 for

non-participants. The higher number of non-participants is to enhance the matching exercise.

However, due to resource constraints, transportation limitations, and busy schedule of some of

the respondents, only 455 was reached among the participants and 522 among the non-

participants, making a total of 977 respondents.


2518-2986 (21-47)

6

Measurement of Youth Empowerment

Youth empowerment was measured with 15 indicators. Based on existing literature on youth

empowerment indicators, a list which describes the important aspects of youth empowerment

was sdeveloped ( Muiruri, 2015 and International Labour Office, 2018).

These indicators are presented in Table 1. The indicators were subjected to Principal Component

Analysis (PCA) in order to extract the essential components required to construct a non-

standardized youth empowerment index (YEI). Based on Kaiser’s criteria (Kaiser, 1960

eigenvalue of 1 and above), several components were generated and retained for the construction

of YEI. The proportion of each of the retained component was used as the weights in generating

the non-standardized index following Equation 1. Also, the KMO value was used to show how

much the retained components explained the variation in the data. According to Antony &

Visweswara (2007), this value is considered as the middling which implies that the data is good

for the analysis.

= {

} (1)

Where:

PC = Principal component

i = Value of retained PC

= Non-standardized Youth Empowerment Index

To standardize the index, Equation 2 was be applied.

=

(2)

Where:

= Standardized Youth Empowerment Index


2518-2986 (21-47)

7

= Non-standardized Youth Empowerment Index

= Minimum non-standardized Youth Empowerment Index

= Maximum non-standardized Youth Empowerment Index

Propensity Score Matching Method

This study adopted the Propensity Score Matching (PSM) method to deal with the problem of

selection bias since participation in any programme may not be random. Many impact evaluation

studies have adopted this method in analysing the impact of interventions or programmes (Asfaw

& Shiferaw, 2010; Ahmed & Haji, 2014; Haji & Legesse, 2017; Balde, at al., 2019).

The method is one of the non-parametric estimation procedures which relaxes the assumptions of

functional forms imposed by parametric models and it is not restricted by distributional

assumptions (El-Shater et al., 2015). It helps in comparing the observed outcomes of the

participants with the outcomes of the non-participants (Heckman, Ichimura, & Todd, 1998). The

basic idea behind this approach is to find in the group of participants, those who have similar

relevant pre-treatment observable characteristics with the non-participants (Haji & Legesse,

2017). PSM essentially estimates youths’ propensity to participate in the programme and it is

commonly estimated using either the Probit of Logit regression model as a function of the

observable characteristics of the youths and then matches youths with similar propensities. The

PSM produces a variable called the propensity score which is the probability that a youth would

participate in the programme and based on the youth’s observable characteristics. We adopted

the three widely used matching methods or algorithm for the purpose of comparison and

robustness (NNM, CBM and KBM).

However, due to unobservable differences between the participants and non-participants of the

programme, the PSM procedure alone may not yield conclusive results (there is overestimation

or underestimation of the outcome) (Balde, et al., 2019). To avoid this, we perform the test of

balancing. We checked the matching quality through visual examination of the PS graph and by

examining the pseudo r-squared, standardized bias, number of matched, high number of

insignificant variables after matching sample and t-test. According to Caliendo & Kopeinig


2518-2986 (21-47)

8

(2008), a good matching quality is depicted by low pseudo r-squared, a mean standardized bias

between 3 and 5 per cent, a large number of insignificant variables after matching and a large

number of matched sample. Thus, the main feature of the matching procedure involves creating

the conditions of a randomized experiment so as to evaluate a causal effect as in a controlled

experiment (Asfaw & Shiferaw, 2010).

Let and

be the outcome variable for participants and non-participants, respectively. The

difference in the outcome between two groups can be calculated using Equation (3)

= -

(3)

Where:

: Outcome of treatment (Youth empowerment of the ith individual, when he/she

participates in the training programme,

: Outcome of the non-participant

: Change in the outcome which can be attributed to training for the ith individual.

Equation 4 can be expressed in causal effect notational form, by assigning =1 as a treatment

variable taking the value 0 if a respondent did not receive treatment (non-participant) and 1

otherwise. Thus, the Average Treatment Effect can be expressed as:

ATE=E ( | )-E(

| ) (4)

Where:

ATE= Average Treatment Effect, the treatment effect on the outcome variables.

E( | ): Average outcome for those who participated in the programme, ( =1).


2518-2986 (21-47)

9

E( | ): Average outcome of those who did not participate in the programme,

( =0).

The ATT for the sample is given by Equation 5:

ATT=E ( -

) = E( =1)-E(

=1) (5)

The NNM estimator was used to pick the comparison group. This method could use a multiple

nearest-neighbours or single nearest-neighbour with the closest propensity score to the

corresponding participant unit (Asfaw & Shiferaw, 2010). The method could also be applied

with or without replacement where the former allows a given non-participants to match with

more than one participants (Adebayo, et al., 2018; Austin, 2014; Asfaw & Shiferaw, 2010;

Caliendo & Kopenig, 2005). To check the robustness of our result, the impact estimate

calculated using the NNM method was compared to the estimates of the CBM and KBM

estimators.

Variables used in the participation and impact empirical models

Participation was measured as a dummy variable which takes the value of one if a youth

participated in the training and zero otherwise. The other outcome variable considered in this

study is the youth empowerment index (YEI). As discussed earlier, the index was computed

using PCA based on existing literature on youth empowerment. A summary of the variables

included in the logit regression (participation model) is presented in Table 2. The variables were

selected based on scanty existing literature on youth participation in agricultural training

programmes and youth empowerment (International Labour Office, 2018; Muiruri, 2015; United

Nation, 2014; Maigida et al., 2013; Meredith, Lucas, Dairaghi, & Ravelli, 2013; Okoli & Okoli,

2013).

Results and discussion

Factors explaining youth participation in agribusiness training in the Fadama GUYS

programme


2518-2986 (21-47)

10

The main results of the factors explaining the probability of participation in the Fadama GUYS

programme (FGP) are presented in Table 3. The pseudo r-squared was 0.26 which shows that

overall, the characteristics of the participants were not so distinct. This low pseudo r-squared is

good for the analysis because, according to Ahmed and Haji (2014), a low pseudo r-squared

within this range is essential as it helps in finding a good match between two groups. In addition,

the likelihood ratio test and the chi-square also reveal that the model is fit for the analysis.

According to the logistic regression results, only three out of the twelve variables included in the

model (Marital status, Head of household years of education and Productive asset index score)

were not statistically significant (Table 3). On one hand, age, years of formal education of the

youth, ownership of agribusiness venture, intention, youth perception about agribusiness training

and perception about agribusiness positively and significantly influenced youth participation in

agribusiness training in the FGP. On the other, gender and household size had a negative and

significant influence on participation in agribusiness training.

Matching Estimator

According to Radicic et al ( 2014), Caliendo & Kopenig (2008), Sianesi (2002), a good

matching estimator must fulfill a number of criteria which include; low pseudo r-squared, large

number of insignificant variables after matching, large number of matched sample size and low

mean SB between 3 and 5 percent. For the purpose of comparison, this study adopted NNM,

KBM and CBM. However, the best matching algorithm based on the analysis was the NNM with

four matches. Before matching, the pseudo r-squared was 0.2567. however after matching, the

pseudo r-squared was reduced to 0.005 (see Table 4). Also, the likelihood ratio after matching

shows that all the regressors in the treatment group were statistically insignificant which implies

that the null hypothesis of joint significance could be rejected (Caliendo & Kopenig, 2005). The

mean bias after matching was reduced to 4.1 percent from the initial value of 38.1 which

indicates that there was an 89 per cent reduction and falls within the recommended mean bias of

between 3-5 per cent.

Matching Participants and Non- Participants


2518-2986 (21-47)

11

The estimated mean, minimum and maximum values of the propensity scores for all sampled

youths were 0.466, 0.010 and 0.971 respectively (Table 5). The corresponding figures for

participants were 0.634, 0.059 and 0.971 respectively while that of the non-participants were

0.319, 0.010 and 0.923. Thus, the common support region lies between 0.059 and 0.971. To

obtain a good quality match between participants and non- participants, the estimated propensity

scores should satisfy the common support condition. Therefore, observations whose estimated

propensity scores were not within the common support region was discarded so as to avoid bad

matches. Subsequently, respondents with estimated propensity scores of less than 0.059 and

greater than 0.971 were not considered for the matching.

Testing for Common Support

Only 36 cases representing four per cent of the entire sample size were lost to common support

restriction using the NNM estimator (Table 6). This indicates that using the NNM estimator, only

36 respondents have a propensity score which lies outside the common support region (between

0.059 and 0.971). These cases were excluded from the analysis to avoid bad matches. However,

there was enough overlap for the matching exercise. Using the other two estimators (KBM and

CM), more cases (55) representing 7 per cent were lost to common support restriction.

In addition, Figures A.1, A.2, A.3 in show the estimated propensity score on the horizontal axis

and their equivalent frequencies on the vertical axis for the three matching estimators. The three

figures validate that very few cases were excluded from the analysis. Also, the propensity score

distribution of the treated and control groups confirms that both groups had enough overlap

which implies that the common support condition was fulfilled.

Testing the balance of Covariates

Among the three matching estimators tested in this study (namely; the Kernel bandwidth, Radius

Caliper and Nearest Neighbour), the NN matching estimator with 4 matches was found to fit the

data best. After choosing the best estimator, we went further to check the balancing of relevant

covariates in the treatment and control groups. This was done using the criteria already specified.

The results of the covariate balances in Table 6 show that after matching the two (treated and

control) groups had close or almost the same matched sample means which was not the case

before matching. For instance, before matching, the mean age of the participants was 27 years


2518-2986 (21-47)

12

while for non-participants was 24 years. However, after matching, the means were 27 years for

both the control and treated groups (see Table 6). This implies that the covariates were well

balanced and the two groups were comparable. Thus, covariates whose differences were

statistically significant before matching became balanced and statistically insignificant after

matching. This is because, according to Sianesi (2002), matching ensures balancing of the

covariates in order to minimize selection bias. For the purpose of comparison, the results of the

covariate balancing test shows that after matching, there was an 89 per cent reduction in the

mean SB using the NNM and CBM estimators while an 87 per cent reduction was achieved

using the KBM estimator (Table 7). This implies that after matching there was no observable

differences in characteristics of the participants and non-participants.

It was also observed that after matching, using NNM and KBM estimators (Tables 6 and 8), all

the variables became statistically insignificant which was not the case before matching, therefore

indicating a good counterfactual. However, using the CBM estimator, one out of the 12 variables

included in the model remained statistically significant (see Table 7).

Testing for Hidden Bias (Sensitivity Analysis)

One limitation of PSM is that it only compares the treated and control units based on observable

characteristic (Asfaw & Shiferaw, 2010). Thus, it does not account for unobserved

characteristics which may likely influence the decision to participate in the programme which

may lead to hidden bias. Therefore, this raises questions about the consistency and robustness of

the PSM estimates.

However, conducting a sensitivity analysis helps to examine whether the inferences about

participation impacts may be questionable due to unobserved characteristics or variables in the

data set. This study adopted the Rosenbaum bounds (rbounds) test to check the sensitivity of the

estimates, testing the null hypothesis that unobserved characteristics have no effect on the

outcome estimate. According to Hujer, et al. (2004), the gamma level, which is the odd ratio of

differential treatment effect due to unobservable characteristics, is reported at the point where 10

per cent level of significance is exceeded.

The results of the sensitivity analysis on hidden bias reported in Table 9 shows that the critical

level of gamma (γ) for the impact of FGP on youth empowerment varies between 2.90 and 2.95.


2518-2986 (21-47)

13

This implies that for the impact estimates to be nullified, the unobservable variable would have

to increase the odds ratio of participation in the programme by up to 195 per cent. We, therefore,

conclude that even large amounts of unobserved heterogeneity would not change the inference

about the impact estimate of agribusiness training in FGP on level of youth empowerment.

Explaining effect of participation in agribusiness training on youth empowerment

After controlling for observable confounding factors, Table 10 shows the results of the Average

Treatment of the Treated (ATT) of the three matching estimators on the outcome variable, youth

empowerment. The impact of youth’s participation in agribusiness training in FGP on their

empowerment is revealed by the difference between the ATT of the control and treated groups.

The results of the NNM, CBM and KBM estimators in Table 10 show that the participants of

FGP had higher empowerment index score than non-participants. The positive difference in ATT

obtained from the three estimators implies that participation in the programme resulted in a

positive increment in the youth empowerment index score. The results further show that the

difference in the ATTs of the three estimators were statistically significant at 1 per cent (Table

10) which agrees with the findings of Judith (2014) and Caliendo & Kopenig (2005) that the

three estimators should give similar results with only slight differences based on the efficiency of

the matching estimator involved. Thus there is a clear demonstration that agribusiness training

had a positive impact on the youth empowerment in the FGP.

Conclusion and Policy Recommendations

This paper has clearly shown that, after controlling for all confounding factors, participation of

youth in agribusiness training has a positive effect on their empowerment. The causal impact

estimation with the PSM showed that participants of the FGP have significantly higher

empowerment index score compared to the non-participants. Also, results from the sensitivity

analysis showed that the impact estimate is robust against any hidden bias. Thus agribusiness

training programmes in Sub-Saharan Africa have the potential to enhance youth empowerment if

they follow the model of FDP. Governments and other stakeholders are therefore encouraged to

initiate or up-scale implementation of such programmes in their countries.


2518-2986 (21-47)

14

The results of this paper generally confirms the potential direct impact of agribusiness training

programmes on improving youth empowerment and thereby improving agriculture and

livelihoods. It is believed that these programmes are likely to inspire young people to engage in

agribusiness and reduce urban-rural migration, idleness and problems associated with drug and

substance abuse among the youth. Further, such programmes would make youth them acquire

relevant skills required in agripreneurship, thereby opening up employment opportunities and

sustainable means to generate income. Thus, governments, development partners and other

stakeholders should consider agribusiness training programmes as one of the priority

interventions and strategies of youth development and empowerment. This would imply

improvement of budget allocations to youth ministries and relevant state departments in order to

fund such initiatives.

Funding

This research was funded by the International Fund for Agricultural Development (IFAD) under

the grant number 2000001374 “Enhancing Capacity to Apply Research Evidence (CARE) in

Policy for Youth Engagement in Agribusiness and Rural Economic Activities in Africa” Project

in the International Institute of Tropical Agriculture (IITA).


2518-2986 (21-47)

15

References

Adebayo, O., Bolarin, O., Oyewale, A., & Kehinde, O. (2018). Impact of irrigation technology

use on crop yield, crop income and household food security in Nigeria: A treatment effect

approach. AIMS Agriculture and Food, 3(2), 154–171.

http://doi.org/10.3934/agrfood.2018.2.154

Ahmed, B., & Haji, J. (2014). Impact of Small-Scale Irrigation on Farm Households’ Technical

Efficiency: The Case of Eastern Oromia, Ethiopia. Asian Journal of Agricultural Biology,

2(4), 224–234.

Ali, M. A., & Salisu, Y. (2019). Women Entrepreneurship and Empowerment Strategy for

National Development. Journal of Economics, Management and Trade, 1–13.

Antony, G. M., & Visweswara, R. K. (2007). A composite index to explain variations in poverty,

health, nutritional status and standard of living: Use of multivariate statistical methods.

Public Health, 121(8), 578–587. http://doi.org/10.1016/J.PUHE.2006.10.018

Asfaw, S., & Shiferaw, B. (2010). Agricultural Technology Adoption and Rural Poverty:

Application of an Endogenous Switching Regression for Selected East African Countries. In

3rd African Association of Agricultural Economists (AAAE) and 48th Agricultural

Economists Association of South Africa (AEASA) Conference, Cape Town, South Africa (pp.

12–23).

Austin, P. C. (2014). A comparison of 12 algorithms for matching on the propensity score.

Statistics in Medicine, 33(6), 1057–1069. http://doi.org/10.1002/sim.6004

Awogbenle, A. C., & Iwuamadi, K. C. (2010). Youth unemployment: Entrepreneurship

development programme as an intervention mechanism. African Journal of Business

Management, 4(6), 831–835.

Balde, B., Diawara, M., Rossignoli, C., & Gasparatos, A. (2019). Smallholder-Based Oil Palm

and Rubber Production in the Forest Region of Guinea: An Exploratory Analysis of

Household Food Security Outcomes. Journal of Agriculture, 9(2), 41.

http://doi.org/10.3390/agriculture9020041

Caliendo, M., & Kopeinig, S. (2008). Some Practical Guidance for the Implementation of

Propensity Score Matching. Journal of Economic Surveys, 22(1), 31–72.

http://doi.org/10.1111/j.1467-6419.2007.00527.x

Caliendo, M., & Kopenig, S. (2005). Some Practical Guidance for the Implementation of

Propensity Score Matching (Discussion Paper Series No. No. 1588). IZA Discussion Paper.

Bonn Germany. Retrieved from

http://ftp.iza.org/dp1588.pdf%5Cnpapers3://publication/uuid/B68613C9-4FF0-4002-B086-


2518-2986 (21-47)

16

60831932B7D5

El-Shater, T., Yigezu, A., Mugera, A., Piggin, C., Haddad, A., Khalil, Y., & Aw-Hassan, A.

(2015). Livelihoods Effects of Zero Tillage among Small and Medium Holder Farmers in

the Developing World. In 89th Annual Conference of the Agricultural Economics Society,

University of Warwick, England (pp. 1–30). Warwick.

Emeh, I., & Eke, J. (2012). Tackling Youth Unemployment in Nigeria: The Lagos State

Development and Empowerment Programmes Initiatives. Afro Asian Journal of Social

Sciences, 3(34), 2229–5313.

Haji, J., & Legesse, B. (2017). Impact of sedentarization program on the livelihood and food

security of Ethiopian pastoralists. Journal of Arid Environments, 136, 45–53.

http://doi.org/10.1016/j.jaridenv.2016.10.007

Heckman, J. J., Ichimura, H., & Todd, P. (1998). Matching as an econometric evaluation

estimator. The Review of Economic Studies, 65(2), 261–294.

Hope, K. R. (2012). Engaging the youth in Kenya: Empowerment, education, and employment.

International Journal of Adolescence and Youth, 17(4), 221–236.

http://doi.org/10.1080/02673843.2012.657657

Hujer, R., Caliendo, M., & Thomsen, S. L. (2004). New evidence on the effects of job creation

schemes in Germany - A matching approach with threefold heterogeneity. Research in

Economics, 58(4), 257–302. http://doi.org/10.1016/j.rie.2004.07.001

Ibrahim, A. I. (2013). Youth empowerment and development drive: An invaluable step to secure

Nigeria’s future prosperity. Unpublished Essay, CIPMN, Nigeria.

International Labour Office (2018). Women’s and Youth Empowerment in Rural Tunisia-An

assessment using the Women’s Empowerment in Agriculture Index (WEAI). Geneva.

Judith, M. (2014). Impact of World Food Programme’s Purchase for Progress Pilot Project on

Farm Incomes in Uasin Gishu and Narok counties, Kenya. University of Nairobi.

Kaiser, H. F. (1960). The Application of Electronic Computers to Factor Analysis.

Educational and Psychological Measurement, 20, 141–151.

http://dx.doi.org/10.1177/001316446002000116

Maigida, J. F., Saba, T. M., & Namkere, J. U. (2013). Entrepreneurial skills in technical vocational

education and training as a strategic approach for achieving youth empowerment in Nigeria.

International Journal of Humanities and Social Science, 3(5), 303–310.

Meredith, K. L., Lucas, B., Dairaghi, J., & Ravelli, P. (2013). Youth Empowerment: The theory

and its implementation.


2518-2986 (21-47)

17

Muiruri, M. P. D. (2015). Mapping of the Honey Value Chain and Analysis of Changes in

Gender Roles and Factors Infuencing Women Empowerment among Beekeepers in Kitui

County, Kenya. University of Nairobi.

Mutuku, C. M. (2011). Youth Perspectives on their Empowerment in Sub-Saharan Africa: The

Case of Kenya. Kent State University.

National Bureau of Statistics (2011). Annual Abstract of Statistics. Retrieved April 15, 2019,

from www.nigerianstat.gov.ng,

National Bureau of Statistics. (2018). Labor Force Statistics Unemployment and

Underemployment Report.

Nigeria Youth Policy Document (2009). Second National Youth Policy. Retrieved from

http://nigeria.unfpa.org/pdf/snyp2009.pdf

Odeh, M. A., & Okoye, C. O. (2014). Poverty reduction policy and youth unemployment in

Nigeria. Public Policy and Administration Research, 3(4), 92–103.

Okoli, D. I., & Okoli, E. I. (2013). Youth Empowerment through Entrepreneurial Development

in Nigeria. Journal of Educational and Social Research, 19, 224–233.

http://doi.org/10.5901/jesr.2013.v3n9p147

Radicic, D., Pugh, G., & Douglas, D. (2014). Evaluating the effectiveness of public support on

inbound open innovation: evidence from Spanish manufacturing SMEs, 44(0), 1–33.

http://doi.org/10.13140/2.1.3714.1768

Sianesi, B. (2002). An Evaluation of the Swedish system of active labour network programmes

in the 1990s. The Review of Economics and Statistics, 86(1), 133–155.

United Nation (2014). Impact of access to financial services, including by highlighting

remittances on development: Economic empowerment of women and youth. In Impact of

Access to Financial Services, including by Highlighting Remittances on Development:

Economic Empowerment of Women and Youth (Vol. 6). Geneva.

United Nations Development Programme (2018). National Human Development Report 2018.

United Nations Development Programme. Abuja, Nigeria.

Yami, M., Feleke, S., Abdoulaye, T., Alene, A., Bamba, Z., & Manyong, V. (2019). African

Rural Youth Engagement in Agribusiness: Achievements, Limitations, and Lessons.

Sustainability, 11(1), 185.


2518-2986 (21-47)

18

TABLES IN THE TEXT

TABLE 1: INDICATORS OF EMPOWERMENT

Six Domains of

Empowerment (6DE)

Indicators Policy issues that are

generally triggered

Access and decision to credit 1. Decision on use of credit

2. Decisions on source of credit

3. Credit repayment ability Economic empowerment

Asset ownership 4. Personal assets

5. Agricultural (Productive) assets

Youth Livelihood 6. Personal living condition (rated

between 1 and 10)

7. Household living condition

8. contribution to HH income

9. life contentment ( rated between 1

and 10) Economic

empowerment, Social

capital, Decision-making

and representation

Financial Freedom 10.Consistent source of income

11. Control over use of income

12. Dependence on family for basic

needs

Group Membership 13. Membership of association

Relationship 14. Closeness to family members

15. Relationship with family

Source: Author’s computation from existing literature on Youth and Women Empowerment


2518-2986 (21-47)

19

TABLE 2: DEFINITION OF VARIABLES

Variable definition and Codes Measurement

Dependent variables

Participation in the training programme Dummy 1 if Yes, 0 if No

Youth Empowerment Index Continuous

Independent variables

Age Continuous

Years of formal education Continuous

Gender Dummy, 1 if male, 0 if female

Household Size Continuous

Intention to start agribusiness Dummy, 1 if Yes, 0 if No

Asset index Continuous

Years of education of Household head Continuous

Agribusiness ownership Dummy, 1 if Yes, 0 if No

Migration from original residence Dummy, 1 if Yes, 0 if No

Perception about training Dummy, 1 if Yes, 0 if No

Perception of Agribusiness Dummy, 1 if Yes, 0 if No

TABLE 3: RESULTS OF THE LOGISTIC REGRESSION MODEL

Variable Coef. S.E z-value M.E

Age 0.110 0.025 4.480*** 0.027

Years of Formal Education of the youth 0.074 0.031 2.350** 0.018

Gender (Male =1) -0.417 0.183 -2.270** -0.104

Marital Status (Married =1) 0.042 0.254 0.170 0.011

Household size -0.067 0.028 -2.360** -0.017

Ownership of Agribusiness (Yes=1) 0.940 0.218 4.310*** 0.231

Migration status (Migrated =1) 0.397 0.182 2.190** 0.098


2518-2986 (21-47)

20

Agribusiness Intention (Yes =1) 0.674 0.072 9.320*** 0.167

Productive asset index score -0.042 0.033 -1.300 -0.011

Perception about training (Positive =1) 0.593 0.296 2.010** 0.142

Perception about agribusiness (Positive =1) 0.821 0.271 3.030*** 0.192

Head of Household Years of schooling 0.013 0.027 0.480 0.003

Constant -7.152 0.802 -8.910***

No of observation = 977

Pseudo r-squared = 0.2567

LR chi2(12) = 346.49

Prob > chi2 = 0.0000

Log likelihood = -501.6626

Source : Stata Output, calculated from Field survey date, 2019

*** p < 0.01, ** p < 0.05, * p < 0.1

TABLE 4: COMPARISON OF THE PERFORMANCE OF MATCHING ESTIMATORS

Matching

Algorithm

Number of

Insignificant

variables after

matching

Pseudo R2 after

matching

Matched

sample size

Mean SB

Nearest Neighbour Matching

1 11 0.012 941 4.9

2 12 0.009 941 4.7

3 12 0.007 941 4.7

4 12 0.005 941 4.1

Kernel Matching

No Bandwidth 12 0.009 941 4.9

0.01 12 0.007 922 4.7

0.05 12 0.009 941 5.3

Caliper Matching

0.01 11 0.009 922 4.1


2518-2986 (21-47)

21

0.05 11 0.012 941 4.9

0.25 11 0.012 941 4.9

Source: survey data (2019)

TABLE 5: MEAN OF ESTIMATED PROPENSITY SCORES

Group Observation Mean Min Max

Total Respondents 977 0.466 0.010 0.971

Participants 455 0.634 0.059 0.971

Non-participants 522 0.319 0.010 0.923



2518-2986 (21-47)

22

TABLE 6: COVARIATE BALANCING TESTS FOR SELECTION BIAS AFTER MATCHING FOR NNM

Variable Unmatched Mean %bias %reduction

bias

t-test p>t

Matched Treated Control

Age Unmatched 27.33 24.33 69.9

10.92 0.000

Matched 26.89 27.13 -5.6 91.9 -0.77 0.440

Years of Formal

Education

Unmatched 14.48 13.77 24.1

3.78 0.000

Matched 14.34 14.49 -5.1 79.0 -0.76 0.445

Gender (Male =1) Unmatched 0.65 0.68 -6.2

-0.96 0.337

Matched 0.67 0.69 -3.3 46.4 -0.48 0.631

Marital Status

(Married =1)

Unmatched 0.40 0.14 59.1 9.32 0.000

Matched 0.35 0.34 2.2 96.2 0.29 0.772

Household size Unmatched 5.63 6.49 -26.1

-4.03 0.000

Matched 5.79 5.99 -6.0 77.1 -1.04 0.297

Ownership of

Agribusiness (Yes=1)

Unmatched 0.43 0.10 82.0

13.00 0.000

Matched 0.39 0.39 1.2 98.6 0.14 0.888

Migration status

(Migrated =1)

Unmatched 0.65 0.63 4.4

0.68 0.498

Matched 0.65 0.68 -5.4 -23.1 -0.79 0.432

Agribusiness Intention

(Yes =1)

Unmatched 3.67 2.55 95.8

14.90 0.000

Matched 3.59 3.56 2.7 97.2 0.41 0.678

Productive asset index

score

Unmatched 4.68 4.35 12.9

1.99 0.046

Matched 4.62 4.47 6.0 53.0 0.89 0.372

Perception about

training (Positive =1)

Unmatched 0.92 0.84 25.6

3.95 0.000

Matched 0.92 0.92 -2.4 90.6 -0.41 0.679

Perception about

agribusiness (Positive

=1)

Unmatched 0.90 0.80 30.6

4.72 0.000

Matched 0.89 0.88 5.6 81.8 0.89 0.372

Head of Household

Years of schooling

Unmatched 14.76 14.02 21.1

3.30 0.001

Matched 14.54 14.65 -3.4 83.9 -0.55 0.584



2518-2986 (21-47)

23

TABLE 7: COVARIATE BALANCING TESTS FOR SELECTION BIAS AFTER MATCHING FOR CBM


Bias

t-test p>t


Age Unmatched 27.33 24.33 69.9

10.92 0.000

Matched 26.79 27.10 -7.3 89.6 -0.97 0.333

Years of Formal

Education

Unmatched 14.48 13.77 24.1

3.78 0.000

Matched 14.36 14.45 -2.9 88.0 -0.42 0.675

Gender

(Male =1)

Unmatched 0.65 0.68 -6.2

-0.96 0.337

Matched 0.67 0.68 -2.1 65.5 -0.30 0.763

Marital Status

(Married =1)

Unmatched 0.40 0.14 59.1

9.32 0.000

Matched 0.34 0.29 12.9 78.2 1.68 0.094

Household size

Unmatched 5.63 6.49 -26.1

-4.03 0.000

Matched 5.82 5.98 -4.7 82.1 -0.77 0.442

Ownership of


Unmatched 0.43 0.10 82.0

13.00 0.000

Matched 0.38 0.38 -0.6 99.3 -0.07 0.942

Migration status

(Migrated =1)

Unmatched 0.65 0.63 4.4

0.68 0.498

Matched 0.66 0.65 2.1 52.0 0.30 0.767

Agribusiness Intention

(Yes =1)

Unmatched 3.67 2.55 95.8

14.90 0.000

Matched 3.57 3.55 1.9 98.0 0.29 0.773

Productive asset index

score

Unmatched 4.68 4.35 12.9

1.99 0.046

Matched 4.60 4.45 5.9 54.4 0.86 0.393

Perception about


Unmatched 0.92 0.84 25.6

3.95 0.000

Matched 0.92 0.91 1.6 93.9 0.25 0.803

Perception about


=1)

Unmatched 0.90 0.80 30.6

4.72 0.000

Matched 0.89 0.88 4.2 86.1 0.66 0.511

Head of Household

Years of schooling

Unmatched 14.76 14.02 21.1

3.30 0.001

Matched 14.52 14.63 -3.2 84.8 -0.49 0.625


2518-2986 (21-47)

24


TABLE 8: COVARIATE BALANCING TESTS FOR SELECTION BIAS AFTER MATCHING FOR KBM


bias

t-test p>t


Age

Unmatched 27.33 24.33 69.9

10.92 0.000

Matched 26.79 27.14 -8.3 88.2 -1.10 0.271

Years of Formal

Education

Unmatched 14.48 13.77 24.1

3.78 0.000

Matched 14.36 14.55 -6.6 72.7 -0.99 0.322

Gender

(Male =1)

Unmatched 0.65 0.68 -6.2

-0.96 0.337

Matched 0.67 0.69 -3.3 45.8 -0.48 0.635

Marital Status

(Married =1)

Unmatched 0.40 0.14 59.1

9.32 0.000

Matched 0.34 0.32 4.5 92.4 0.58 0.565

Household size

Unmatched 5.63 6.49 -26.1

-4.03 0.000

Matched 5.82 6.03 -6.1 76.6 -1.03 0.304

Ownership of


Unmatched 0.43 0.10 82.0

13.00 0.000

Matched 0.38 0.36 3.7 95.5 0.44 0.661

Migration status

(Migrated =1)

Unmatched 0.65 0.63 4.4

0.68 0.498

Matched 0.66 0.67 -3.3 24.7 -0.47 0.639

Agribusiness

Intention (Yes =1)

Unmatched 3.67 2.55 95.8

14.90 0.000

Matched 3.57 3.52 4.2 95.6 0.64 0.526

Productive asset

index score

Unmatched 4.68 4.35 12.9

1.99 0.046

Matched 4.60 4.41 7.2 44.2 1.04 0.297

Perception about


Unmatched 0.92 0.84 25.6

3.95 0.000

Matched 0.92 0.92 -3.1 88.0 -0.51 0.608

Perception about


=1)

Unmatched 0.90 0.80 30.6

4.72 0.000

Matched 0.89 0.88 2.6 91.4 0.41 0.679

Head of Household Unmatched 14.76 14.02 21.1

3.30 0.001


2518-2986 (21-47)

25

Years of schooling Matched 14.52 14.66 -3.9 81.4 -0.63 0.531

TABLE 9: INDICATORS OF MATCHING QUALITY AND ROBUSTNESS OF

RESULTS


TABLE 10: IMPACT OF AGRIBUSINESS TRAINING ON YOUTH EMPOWERMENT IN FGP


FIGURE A. 1: COMMON SUPPORT GRAPH FOR NNM ALGORITHM

0 .2 .4 .6 .8 1Propensity Score

Untreated Treated: On support

Treated: Off support

Mean bias

Pseudo-R2 p>chi2 Gamma

Matching

algorithms

Before After % bias

reduction

Unmatched Matched Unmatched Matched

NNM 38.1 4.1 89 0.26 0.005 0.000 0.930 2.90-2.95

KBM 38.1 4.7 87 0.26 0.007 0.000 0.840 2.90-2.95

CBM 38.1 4.1 89 0.26 0.009 0.000 0.597 2.90-2.95

Outcome

variable

Matching

algorithm

Treated Control ATT

(Difference)

Std.

Error

t-value

Youth

Empowerment

Index

NNM 6.416 5.692

0.724 0.116 6.25***

CBM 6.393

5.557 0.836 0.134 6.24***

KBM 6.393 5.608 0.784 0.112 7.03***


2518-2986 (21-47)

26

Source: Stata Output, generated from Field Survey data using Psgraph, 2019

FIGURE A. 2: COMMON SUPPORT GRAPH FOR KBM ALGORITHM

Source : Stata Output, generated from Field Survey data using Psgraph, 2019

FIGURE A. 3: COMMON SUPPORT GRAPH FOR CBM ALGORITHM

Source: Stata Output, generated from Field Survey data using Psgraph, 2019







african multidisciplinary journal of research (amjr) vol

Documents